JPS6335071A - Variable power method - Google Patents

Abstract

PURPOSE:To reserve the information of a detailed part of an image in a state that a halftone reproducibility remains maintained as it is, by detecting whether density information and edge information exist or not by a small block unit, from a binary original image, and switching adaptively a variable power processing method in accordance with the presence and absence of the edge information. CONSTITUTION:As for an image whose density variation is flat, the deterioration of a halftone expressiveness is related to the deterioration of the picture quality as a visual sense, and on the contrary, as for an image containing many edge information, the deterioration of a resolving power is related to the deterioration of the picture quality. Accordingly, by switching these two variable power methods in accordance with the quantity of edge information, the deterioration of the picture quality is reduced. A binary data outputted from a dither processing circuit 2 and a line buffer 7 is inputted to an edge information detecting circuit 8. An area deciding circuit 9 adds the edge information of a 4-line portion, and outputs a select signal 1, when the edge information exceeds 4 picture elements at every 4X4 areas. The result of decision outputted from the area deciding circuit 9 is inputted to a selector 6, and uses as a select signal for switching adaptively variable power images of 2 systems from a dither circuit 3 and a variable power circuit 5.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field) The present invention relates to a method of processing an image binarized by area gradation method (dither processing), and particularly to a scaling method thereof.

(Prior art) In order to express a half-tone image like a photograph using a dot printer such as a laser printer, thermal transfer printer, or inkjet I-printer, an areal scale that changes the number of dots per unit area is used. A formula is used. At this time, in consideration of gradation and low noise, it is common to use a systematic dither method.

Incidentally, an advantage of digitizing images is that data can be stored, transmitted, edited, and the like. Now, considering the amount of image data, if an A4 size original is read as multi-gradation data of 8 bits per pixel at a density of 4QQdpi, it will be about 15 megabytes.

On the other hand, image data that has been binarized for printer output has 1 bit per pixel, so per A4 sheet,
It becomes 1.9 megabytes. As described above, image data generally has a huge amount, so when considering data storage and transmission, handling multi-level data as is is extremely expensive and unrealistic.

Therefore, halftone image data is stored as binary data. Furthermore, when editing an image, there may be times when it is desired to perform enlargement or reduction processing on the saved data, and in this case, scaling processing of the binary image is required.

Various methods have been proposed for enlarging and reducing binary images.

For example, a processing method is known in which an enlarged image as shown in FIG. 2 is obtained by enlarging each bit data of the original image shown in FIG. 1 twice in both the vertical and horizontal directions.

When reducing an image, for example, by sampling the bit data of the original image shown in Fig. 1 every other bit in both the vertical and horizontal directions, a 1/2 reduced image as shown in Fig. 3 is obtained. Processing methods are known.

This method does not pose any major problems when applied to normal binary images, but when applied to dithered images, the dither pattern is destroyed, resulting in poor image quality after enlargement or miniaturization. There was a risk of deterioration.

In the techniques disclosed in Japanese Unexamined Patent Publication Nos. 61-35070 to 35073, magnification processing of binary images is performed by the following method.

(1) Count the number of black dots for each unit area and obtain average density data.

(2) Perform scaling processing on density data, that is, multivalued data.

(3) The multivalued image data after scaling is converted into two-digit data by dither processing.

In this method, binary data is converted to density data and scaling processing is performed, so that the halftone expressivity does not deteriorate. However, 2
In order to obtain density (shade) information from value image data, the dot density (=average density) of a unit area is detected. With this method alone, for example, if the unit area is 4 x 4 (dot * dot) and the reading density is 400 dpi, 10
Since frequency components of 0 dots/inch = 4 lines/mm or more disappear, the image becomes a so-called blurred image with low resolution, especially during enlargement.

(Objective) The object of the present invention is to provide a magnification change method that eliminates the drawbacks of the above-mentioned prior art and can preserve information on detailed parts of an image, especially during enlargement, while maintaining halftone reproducibility. There is a particular thing.

(Configuration) The algorithm of the present invention detects the presence or absence of density information and edge information in small block units from a binary original image, and adaptively switches the scaling processing method depending on the presence or absence of edge information. .

First, the scaling algorithm used in the present invention will be explained with reference to the drawings.

FIG. 4 shows a part of a halftone original and an example of sampling by a reading scanner. In the present invention, characters mixed in a halftone image, ! The purpose is to save edge information such as #ii. Therefore, the illustrated manuscript assumes some characters in the middle "1546M area."

FIG. 5 shows a reading date 76 of the document shown in FIG. However, since it is single-sided, the number of gradations is 16 steps [0 (white) to I
5BB> ). When this multivalued image data is binarized using the dither matrix shown in FIG. 11, it becomes as shown in FIG. The image lacks sharpness compared to the original. Generally, when dither processing is performed, the transmission of high spatial frequency components is poor, resulting in a decrease in resolution. In order to correct this decrease in resolution, for example, a digital filter as shown in Fig. 7 is used,
There is a method of performing so-called edge enhancement processing.

FIG. 6 shows multivalued image data that has been passed through the edge enhancement filter shown in FIG. If this is binarized using the dither matrix shown in FIG. 11 as described above, the result will be as shown in FIG. 9. ] In the scaling process below, the 2
Let us use the value image as the original image.

As mentioned above, the algorithm of the present invention detects the presence or absence of 4-degree information and edge information in small blocks from a binary original image, and adaptively switches the variable Isa processing method depending on whether or not edge cleaning is performed. be.

FIG. 10 shows average density data for a 4×1 area obtained by counting the number of black dots in the area.

Next, a method for detecting Edge+:1 report will be explained.

First, the multivalued data shown in FIG. 10 is subjected to binary processing using the dither pattern used to obtain the original image, that is, the dither matrix shown in FIG. 1I. The result is the 12th
As shown in the figure. If we compare Figure 9 and Figure 12 here, we can see that
In the figure, since average density data is used, high spatial frequency components are lost from the original image, resulting in a blurred image with no edge information. Therefore, the ninth
By subtracting the blurred image in Figure 12 from the original image in the figure,
Edge information can be extracted. FIG. 13 shows the edge information thus obtained. Here, the circle mark indicates recorded (black dot) information, and the X mark indicates non-recorded (white dot) information.

Although FIG. 9 assumes an area having edge information such as characters, conversely, consider an area where the density changes gradually. FIG. 14 assumes multivalued data obtained by reading a region with a gentle density change using a scanner. FIG. 15 shows data obtained by passing this image data through the digital filter shown in FIG. 7 in the same way as the image of the edge area.

Further, FIG. 16 shows an image binarized using the dither pattern shown in FIG. 11. This binary image will be used as the original image of the flat area. Edge information is detected using the method described above. Figure 17 is the average concentration data,
FIG. 18 is the dithered image. By subtracting the blurred image shown in FIG. 18 from the original image shown in FIG. 16, edge information shown in FIG. 19 is obtained.

Comparing the amount of edge information detected from FIG. 9 and FIG. 16, that is, the number of pixels marked with ○ or × in FIG. 13 and FIG. (=4×4), there are 0 to 2 pixels in a flat area with 4 to 10 pixels. In other words, the degree of coincidence between the original image and the blurred image is good in flat areas, and the degree of coincidence between the two is poor in areas that include text. A flat area is an area where there are few components with high spatial frequencies, so less information is lost due to averaging, which explains why the original image and the blurred image match well.

In reality, even areas with flat density may contain high-frequency noise, and this tendency becomes even stronger when edge enhancement processing is performed. As a countermeasure, it is sufficient to set an appropriate dark value for the amount of edge information when determining whether or not it is an edge area.

In this example, when there are four or more edge pixels per unit area, it is determined that the area is an edge area, and when there are three or less edge pixels, it is determined that the area is a flat area.

Next, a method of scaling a binary image will be described.

FIG. 20 shows the average density in 4×4 pixel units obtained from the image shown in FIG. This is a binary enlarged image. Similarly, if you enlarge the image in Figure 16,
The image shown in FIG. 21 is obtained. Although this method has good halftone expression, as can be seen from FIG. 20, the resolution is significantly reduced. When changing the magnification, some improvement is possible by using an interpolation method such as polynomial approximation, but there is a limit.

Figures 22 and 23 are similar to Figures 9 and 1, respectively.
This is an image obtained by simply enlarging one pixel of the binary image shown in FIG. 6 to 2×2 pixels using the method described in the prior art section. This scaling method is a method with excellent resolution because edge information is preserved as is, but the dither pattern may become rough or destroyed, resulting in poor halftone expressivity.

Considering deterioration in image quality here, in an image with flat density changes as shown in FIG. 16, deterioration in halftone expression visually leads to deterioration in image quality. On the other hand, in an image containing a lot of edge information as shown in FIG. 9, a decrease in resolution leads to a deterioration in image quality. Therefore, it can be seen that deterioration in image quality can be reduced by switching between these two scaling methods depending on the amount of edge information.

By the way, this method requires data on the dither pattern used to obtain the binary original image and information on the phase difference between the image data and the dither matrix. This can be accomplished as follows. It is assumed that the binary image data obtained in this process is stored in an external storage device such as a magnetic disk, magnetic tape, or optical disk, or is transmitted via communication. Therefore, when storing and transmitting image data, data regarding the dither pattern may be attached to the image data as a header. When performing scaling processing, a part of the header of the image data to be processed is decoded, and the size of the dither matrix, darkness value, etc.
Try to obtain a phase difference with the image. If several fixed dither patterns are used in a series of systems, the data added as a header at this time may be only the serial number and phase information of the dither patterns. Also,
- For boat fishing, images that express halftones have poor compression efficiency.
Considering that the amount of data is extremely large, even if all the dark value data of the dither pattern is attached to the header, it will not be a big burden, but rather the dither pattern data stored in a large number of ROMs etc. in the variable magnification device. It also has the advantage of saving storage space.

FIG. 24 shows an example of an image processing apparatus for performing magnification processing according to the method of the present invention. 1 is a circuit that detects the average density in 4×4 block units from binary data. This circuit consists of a ROM for counting the number of recorded pixels (black pixels) in binary data, an adder for adding four lines of data, a RAM for holding the added data, etc. ing. Reference numeral 2 denotes a circuit that binarizes the density data from the density data detection circuit 1 by dither processing, and uses a ROM or RAM. When a RAM is used, dither pattern data is externally written according to the header information of the original image. In both ROM and RAM, binary data can be output in a table-reference manner by addressing with multi-level density data and vertical and horizontal address data. 7 is a line buffer for synchronizing the binary data of the blurred image output from the dither processing circuit 2 and the binary data of the original image. Binary data output from the dither processing circuit 2 and line buffer 7 is input to an edge information detection circuit 8. This uses a comparator,
When the two binary data values match, O (non-edge pixel) is output, and when they do not match, 1 (edge pixel) is output. The output of the edge information detection circuit 8 is input to the area determination circuit 9. The area determination circuit 9 adds edge information for four lines,
When the edge information is 4 or more pixels for each 4×4 area, select signal 1 is output. For example, the area determination circuit 9
As shown in FIG. 5, it consists of a 3-line buffer, an adder, and a line buffer for holding area determination results and synchronizing with variable-scale image data output from a dither circuit 3 and a variable-magnification circuit 5, which will be described later. The determination result output from the area determination circuit 9 is input to the selector 6 and is used as a selection signal for adaptively switching between the two systems of variable magnification images from the dither circuit 3 and the variable magnification circuit 5. Reference numeral 4 denotes a circuit that scales the density data, and performs processing such as interpolation and thinning according to the enlargement and reduction ratio. The density data scaled by the scaling circuit 4 is binarized by the dither circuit 3. Dither circuit 3
is similar to dither processing circuit 2, but the dither pattern may be the same as or different from 2. 5 is a 2(i data scaling circuit), which has a line buffer to synchronize with the binary image from the dither circuit 3. The binary image output from the dither circuit 3 and the scaling circuit 5 is As described above, the image data from 5 is selected in the case of an edge area, and the image data from 3 is selected in the case of a non-edge area, depending on the selection signal indicating whether the area is an edge area or not.

(Effects) As explained above, with the processing method of the present invention, it is possible to perform region determination on the original image and adaptively switch the magnification processing that prioritizes gradation or resolution. A variable magnification image with good gradation and resolution can be obtained.

[Brief explanation of drawings]

Figure 1 is a diagram showing each bit data of the original image, Figure 2 is a diagram showing an enlarged image based on Figure 1, Figure 3 is a diagram showing the same reduced image, and Figure 4 is a diagram of a halftone original. FIG. 5 is a diagram showing the read data of the original, FIG. 6 is a diagram showing multivalued image data passed through an edge enhancement filter, and FIG.
The figure shows a digital filter pattern, Figure 8 shows a pattern obtained by converting the data shown in Figure 5 into a binary format using a dither matrix, and Figure 9 shows the data shown in Figure 6 converted into a binary pattern using a dither matrix. A diagram showing a pattern corrected by using
Figure 10 is a diagram showing density information in small block units;
The figure shows a dither matrix, FIG. 12 shows a pattern obtained by binarizing the multivalued data in FIG. 10 using a dither matrix, FIG. 13 shows edge information, and FIG.
The figure shows multi-value data in a region with a gentle concentration change.
Figure 15 is a diagram showing the data obtained by applying a digital filter to this data, Figure 16 is a diagram showing an image pattern further binarized using a dither pattern, and Figure 17 is a diagram showing the image pattern 10
Figure 18 is a diagram showing the same dithered image; Figure 19 is a diagram showing edge information similar to Figure 13; Figures 20 and 21 are similar to the same magnification method. FIG. 22 and FIG. 23 are diagrams showing enlarged image patterns obtained by a common scaling method different from the above. FIG. 24 is a block diagram showing an example of a scaling processing circuit. FIG. 25 is a block diagram showing the inside of the main part thereof. Figure 1 Tree l Figure 3 Figure 2 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16 Figure 17 Figure 18 Figure 19 Figure 20 Figure 21

Claims (1)

[Claims] In an image processing method that performs scaling processing on an image representing halftone density using the area gradation method, the presence or absence of density information and edge information is detected in small blocks from a binary original image, and the presence or absence of edge information is detected. A magnification changing method characterized in that one of at least two different types of magnification changing means is selected depending on the size of the image.